A low-complex multi-channel methodology for noise detection in phonocardiogram signals

Area of Science:

• Biomedical Engineering
• Signal Processing
• Cardiology

Background:

• Phonocardiography (PCG) is vital for diagnosing heart conditions.
• PCG signals are susceptible to noise, hindering accurate heart sound interpretation, especially in p-health settings.
• Noise interference, often overlapping with PCG frequencies, complicates analysis.

Purpose of the Study:

• To develop and evaluate a method for detecting noisy periods within phonocardiography (PCG) signals.
• To classify signal content on a window-by-window basis, differentiating clean from contaminated PCG sections.
• To address the challenge of noise in PCG analysis for improved diagnostic accuracy.

Main Methods:

• A two-phase algorithm was developed for PCG noise detection.
• Phase one identifies noise-free windows using time-domain features.
• Phase two compares noise-free windows with the signal using frequency-domain features for classification.

Main Results:

• The developed algorithm successfully discriminated between clean and contaminated PCG segments.
• Achieved an average sensitivity of 95.59% in detecting noisy sections.
• Achieved an average specificity of 92.68% in identifying clean sections.

Conclusions:

• The proposed method effectively detects noise in PCG signals, enhancing diagnostic reliability.
• This technique is valuable for accurate heart sound analysis in p-health environments.
• The algorithm's high sensitivity and specificity demonstrate its potential for clinical application.